INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ornia

-0.80

wagen

-0.72

 Rite

-0.70

aday

-0.69

 Galile

-0.68

issan

-0.63

SPONSORED

-0.63

 monks

-0.59

Huh

-0.58

 Chop

-0.58

POSITIVE LOGITS

urated

0.76

icator

0.68

ENDED

0.67

itious

0.66

Either

0.64

ALS

0.64

itably

0.64

ãĤ¤ãĥĪ

0.64

ailability

0.64

Lost

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.